Sponsored by:
IEEE TCCA, IEEE TCPP, ACM SIGARCH, & IFIP Working Group 10.3
& Generous Corporate Support from
Intel, SGI, & IBM
e-mail:
Keynote Speaker: Michael D. Smith, Division of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138.
Title: Dynamic Optimization: An Online Opportunity
Abstract: Dynamic optimization, the run-time modification or rearrangement of the dynamic instruction stream, is a well-accepted technique used by almost every hardware manufacturer to improve the performance of the applications running on their processors. With the growing popularity of systems for binary translation and virtual machines for "write once, run anywhere" code, we are beginning to see the benefits of having a software layer above the hardware that aids in the dynamic translation and optimization. Today, manufacturers like Transmeta with their Code Morphing software, Sun with their Java HotSpot optimizer, and HP with their Dynamo system hide the power of this layer in the name of transparency. What if the software layer in these systems were not invisible? What further benefits could we reap? Looking deeper, what kinds of interfaces might we export and how might we best organize such a system? This talk explores some answers to these and other similar questions.
Bio: Michael D. Smith is Gordon McKay Professor of Computer Science and Electrical Engineering in the Division of Engineering and Applied Sciences at Harvard University. His research interests lie in computer architecture and back-end compilation, and his current projects focus on techniques for feedback-directed code optimization. His research group distributes an experimental compiler called Machine SUIF, which is a component of the National Compiler Infrastructure project sponsored by DARPA and NSF. Mike received his Ph.D. in Electrical Engineering from Stanford University in 1993. Please visit http://www.eecs.harvard.edu/smith for more information.
Questions or comments, please contact the Webmaster.